Messages can be transmitted between subscriber units (computers), arranged in a single network, by various methods.
In a conventional, generally known "point-to-point" method for transmitting messages, messages are directly transmitted from one subscriber unit, a message sender, to another subscriber unit, a message recipient. This method provides the most economical way for transmitting messages. The method, however, suffers a significant problem: it is entirely impossible to transmit a message to a disabled destination subscriber unit, i.e. when the destination subscriber unit is turned off at the moment of transmitting a message to it and will be subsequently turned on after a lapse of time (off-line mode).
When a central server is used, a message sent by one network subscriber (a sender) to another subscriber (a called party) is first transmitted to a network central service unit (a server). The destination unit either periodically checks the central server for messages that are intended for it, or is informed of the presence of such messages by the central server itself, and receives the messages. This method allows the messages to be transmitted in the off-line mode.
The problem with the method for transmitting messages through a central server is the fact that a message transmission rate depends on the central server efficiency and a total number of simultaneously accessing subscribers. If the number of simultaneously operating subscribers is significant, heavy demands are imposed on the central server efficiency.
To improve the system reliability and to provide a possibility to occasionally disable the central server without disturbing the normal operation, the central server can be complemented with one or more backup servers. Messages transmitted via the central server are also simultaneously duplicated at the backup servers. When the central server is turned off, it will be replaced by a backup server, and, therefore, the system will not be disordered even if the central server is disabled.
A disadvantage of the redundant system is a necessity to duplicate operations of the central server at least at one backup server, i.e. to invoke additional resources. In this method, the message transmission rate is also defined by the relationship between the central server efficiency and a total number of simultaneously operating subscribers.
Known in the art is a method for transmitting messages, utilized in the Internet, that is based on a static routing principle. In this method of transmitting messages, each server having message reception and transmission functions, i.e. a "mail" server, comprises a list of other "mail" servers to which mail messages intended for destinations that are external to said server can be sent. If it is necessary to send a message to a destination unit that is assigned to another mail server, the next location to store the message will be selected from the static list of mail servers (see Mail Routing and the Domain System. C. Partridge. CSNET CIC BBN Laboratories Inc., January 1986. Network Working Group. Request for Comments: 974).
However, the method based on static routing fails to take into account such factors as a traffic load on servers. A route should be changed either through a direct instruction from the sender, or by a system administrator reconfiguring the mail server (see Classification in E-mail Routing. J. Houttuin, RARE October 1994, Network Working Group. Request for comments: 1711).